Categorization

1
Chapter 8

• Categorization

2
Humans have a tendency to categorize
3
Benefits of categorizing

• Reduces complexity of environment
• Means by which objects are identified
• Reduces the need for constant learning
• Indicates appropriate behaviors
• Relates classes of objects and events

4
Categories represent concepts.
5
Categories represent concepts. A concept can be
defined as any regularity among a set of objects
or events.
6
Early studies of categorization used the concept
identification paradigm.
7
(No Transcript)
8
Participants must discover the rule that leads to
correct classification of the stimuli.
9
Criticism of the method
The categories (concepts) are artificial and lack
some of the characteristics of real-life
(natural) categories.
10
Characteristics of real life categories

• Real-life categories often have fuzzy boundaries.
• Members of a real-life category are not always
  equal.
• There are different levels of categorization in
  real life

11
Where is the boundary between red and orange?
12
Superordinate Category
Level 1
Level 2
Level 2
Level 3
Level 3
Level 3
Level 3
Subordinate Category
13
Animal
Dog
Cat
Poodle
Collie
Siamese
Persian
14
Basic Level Categories
Dog and cat are both examples of basic level
categories.
15
Basic Level Categories
Dog and cat are both examples of basic level
categories. A basic level category is the level
of categorization at which the greatest number of
distinguishing features correlate.
16
Example
What are the distinguishing features of the
following categories?
Furniture
Table
Chair
Kitchen Table
Dining Table
Desk Chair
High Chair
17
Evidence in support of basic level categories

• Children learn to classify objects according to
  basic level categories first.
• People are faster when verifying that an object
  belongs to a basic level category.
• Basic level categories seem to be universal.

18
Theories of categorization

• Prototype theory
• Feature theory
• Exemplar theory

19
Prototype theory

• When classifying something, the individual
  compares the item to stored prototypes.
• A prototype can be thought of as the best, most
  representative member of a category. It often is
  defined as the statistical average for the
  category
• The prototype which provides the closest match
  determines the category to which the item
  belongs.

20
Empirical Support

• Participants learn to classify dot patterns
• The different categories are based on different
  prototypes
• The participant never sees the prototype during
  the learning phase
• During the test phase, the prototype and new
  variations are presented for classification
• Results

21
(No Transcript)
22
Prototype
23
Variations constructed by moving dots
24
A second variation
25
Some variations involve more than one
transformation
26
(No Transcript)
27
(No Transcript)
28
Feature Theory
Category membership is determined by the features
that are present.
29
Example
The category bird has the following features

• Feathers
• Wings
• Can fly
• Builds nests
• Lays eggs

30
Example
A robin would be classified as a bird since it
shares the same features
Bird
Robin

• Feathers
• Wings
• Can fly
• Builds nests
• Lays eggs

• Feathers
• Wings
• Can fly
• Builds nests
• Lays eggs
• Red breast

31
Some feature theories distinguish between
defining and characteristic features.
Feathers Wings Lays eggs Builds nests Can fly
Defining
Characteristic
32
Some problems for feature theory

• Although someone may be able to reliably
  categorize a set of objects, they may not be able
  to identify the defining features
• Eliminating a defining feature doesnt
  necessarily change the classification.

33
Exemplar Theory
The basic notion of exemplar theory is that we
dont store knowledge. Rather we store specific
experience.
34
Exemplar Theory
The basic notion of exemplar theory is that we
dont store knowledge. Rather we store specific
experience. In essence, the argument dispenses
with the distinction between episodic and
semantic memory and asserts that there is only
episodic memory
35
Some support comes from the observation that
categorization can be influenced by context.
Recall that episodic memory is supposed to be
influenced by context, but semantic memory is not.
36
Which are cups and which are bowls?
37
Which are cups and which are bowls?
Would it make a difference if you imagined them
with filled with mashed potatoes?
38
Data
39
Classification can also be influenced by a single
episode
40
The fact that classification can be influenced by
context, and more importantly, by a single
episode is taken as evidence which is supportive
of exemplar theory.